Remote check-in method and system based on large language model and storage medium

Abstract

The invention provides a remote check-in method and system based on a large language model and a storage medium, and the method comprises the steps: inputting check-in interaction information into a pre-trained large oracle model for semantic analysis, and obtaining interaction semantics; performing interaction feedback on the user according to the interaction semantics; if the interaction semantics comprise preset remote check-in semantics, check-in preferences are determined according to historical check-in information; according to the check-in preference, the travel destination and the check-in demand, check-in recommendation information is determined, and remote check-in recommendation is carried out on the user according to the check-in recommendation information; and performing remote check-in registration on the user according to the remote check-in recommendation. According to the embodiment of the invention, remote check-in recommendation can be effectively carried out on the user based on the check-in habit of the user through the check-in recommendation information, the remote check-in operation of the user is effectively facilitated, and the use experience of the user is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence, and particularly to a remote check-in method, system and storage medium based on a large language model. Background Art

[0002] With the growth of air passenger traffic and the increasing demand for service convenience by passengers, on-site check-in at airports faces problems such as long queuing times and unreasonable resource allocation. Therefore, remote check-in methods have received increasing attention.

[0003] In the existing remote check-in process, generally, personal information is input through the official website of the airline and a flight is selected for check-in. The user operation is cumbersome and complex, reducing the user experience. Summary of the Invention

[0004] The purpose of the embodiments of the present invention is to provide a remote check-in method, system and storage medium based on a large language model to solve the problem of cumbersome and complex remote check-in operations in the prior art.

[0005] The embodiments of the present invention are implemented as follows. A remote check-in method based on a large language model, the method includes:

[0006] Obtain the user's check-in interaction information, and input the check-in interaction information into a pre-trained large prediction model for semantic analysis to obtain interaction semantics;

[0007] Determine an interaction reply according to the interaction semantics, and perform interaction feedback on the user according to the interaction reply;

[0008] If the interaction semantics contain preset remote check-in semantics, obtain the travel destination and check-in requirements in the interaction semantics;

[0009] Obtain the user's historical check-in information, and determine the check-in preference according to the historical check-in information;

[0010] Determine check-in recommendation information according to the check-in preference, the travel destination and the check-in requirements, and perform remote check-in recommendation on the user according to the check-in recommendation information;

[0011] If a confirmation message for the remote check-in recommendation is received, perform remote check-in registration on the user according to the remote check-in recommendation.

[0012] Preferably, determining the check-in preference according to the historical check-in information includes:

[0013] Obtain the historical flight and historical seat in the historical check-in information, and determine the flight preference according to the flight information of the historical flight;

[0014] Determine seat preferences based on the seat information of the historical seats, wherein the check-in preferences include the flight preferences and the seat preferences.

[0015] Preferably, determine check-in recommendation information according to the check-in preferences, the travel destination, and the check-in requirements, including:

[0016] Obtain the departure date in the check-in requirements, and determine the recommended flight number according to the departure date, the travel destination, and the flight preferences;

[0017] Obtain the flight ticket sales information of the recommended flight number, and determine the recommended seat according to the flight ticket sales information and the seat preferences;

[0018] Generate the check-in recommendation information according to the flight information of the recommended flight number and the recommended seat.

[0019] Preferably, after determining the recommended seat according to the flight ticket sales information and the seat preferences, it further includes:

[0020] Obtain the historical travel mode and historical waiting time in the historical check-in information, and determine the travel preference according to the historical travel mode;

[0021] Obtain the current accommodation location of the user and the airport location corresponding to the recommended flight number, and determine the recommended departure time according to the travel preference, the current accommodation location, the airport location, and the historical waiting time;

[0022] Obtain the accommodation information in the historical check-in information, and determine the accommodation preference according to the accommodation information;

[0023] Determine the accommodation recommendation according to the travel destination and the accommodation preference, and generate the check-in recommendation information according to the flight information of the recommended flight number, the recommended seat, the accommodation recommendation, and the recommended waiting time.

[0024] Preferably, after remotely recommending the check-in recommendation information to the user, it further includes:

[0025] If an artificial assistance instruction for the remote check-in recommendation is received, connect the user with the remote check-in customer service for audio and video, and obtain the remote check-in interaction information of the user within a preset duration;

[0026] Send the remote check-in interaction information and the check-in recommendation information to the remote check-in customer service, and obtain the artificial consultation questions of the user in real time;

[0027] Determine the question and answer reply according to the artificial consultation questions, and send the question and answer reply to the remote check-in customer service.

[0028] Preferably, before inputting the check-in interaction information into a pre-trained large prediction model for semantic analysis to obtain interaction semantics, it further includes:

[0029] Obtain check-in interaction samples, and input the check-in interaction samples into the large prediction model for word embedding processing to obtain word embedding features;

[0030] Obtain the context features of the check-in interaction samples, and fuse the context features and the word embedding features to obtain fused features;

[0031] Perform semantic prediction on the fused features to obtain predicted sample semantics, and determine a predicted sample response according to the predicted sample semantics;

[0032] Determine a model loss according to the predicted sample response, the predicted sample semantics, and the word embedding features, and update the large language model according to the model loss until the large language model converges to obtain the pre-trained large prediction model.

[0033] Preferably, obtaining the context features of the check-in interaction samples includes:

[0034] Obtain the hidden state of the current text input and the hidden state of the previous text in the check-in interaction sample, and determine a forgetting coefficient according to the hidden state of the current text input and the hidden state of the previous text;

[0035] Determine forgetting information according to the forgetting coefficient and the memory state of the previous text, and perform a non-linear transformation on the hidden state of the current text input and the hidden state of the previous text to obtain an input coefficient;

[0036] Update the memory state of the current text input according to the input coefficient, and determine an output gate vector according to the hidden state of the current text input and the hidden state of the previous text;

[0037] Determine the context features of the current text input according to the output gate vector and the updated memory state of the current text input.

[0038] Another object of the embodiments of the present invention is to provide a remote check-in system based on a large language model, and the system includes:

[0039] A semantic analysis module, configured to obtain the check-in interaction information of a user, and input the check-in interaction information into a pre-trained large prediction model for semantic analysis to obtain interaction semantics;

[0040] An interaction feedback module, configured to determine an interaction response according to the interaction semantics, and perform interaction feedback on the user according to the interaction response;

[0041] A requirement acquisition module, configured to acquire the travel destination and check-in requirements in the interaction semantics if the interaction semantics include a preset remote check-in semantics;

[0042] A preference determination module, configured to acquire the user's historical check-in information, and determine check-in preferences according to the historical check-in information;

[0043] A check-in recommendation module, configured to determine check-in recommendation information according to the check-in preferences, the travel destination, and the check-in requirements, and perform remote check-in recommendation on the user with the check-in recommendation information;

[0044] If a confirmation message for the remote check-in recommendation is received, perform remote check-in registration on the user according to the remote check-in recommendation.

[0045] Preferably, the preference determination module is further configured to: acquire the historical flight and historical seat in the historical check-in information, and determine flight preferences according to the flight information of the historical flight;

[0046] Determine seat preferences according to the seat information of the historical seat, where the check-in preferences include the flight preferences and the seat preferences.

[0047] In an embodiment of the present invention, by inputting check-in interaction information into a pre-trained large prediction model for semantic analysis, the interaction semantics of the user can be effectively acquired. Through the interaction semantics, the interaction response can be effectively determined. Through the interaction response, interaction feedback can be effectively performed on the user. By acquiring the user's historical check-in information, the check-in preferences of the user can be effectively determined. Based on the check-in preferences, travel destination, and check-in requirements, the check-in recommendation information can be automatically determined. Through the check-in recommendation information, remote check-in recommendation can be effectively performed on the user based on the user's check-in habits, effectively facilitating the user's remote check-in operation and improving the user's experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a flowchart of a remote check-in method based on a large language model provided in the first embodiment of the present invention;

[0049] Figure 2 is a schematic structural diagram of a remote check-in system based on a large language model provided in the second embodiment of the present invention;

[0050] Figure 3 is a schematic structural diagram of a terminal device provided in the third embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0052] In order to illustrate the technical solution described in the present invention, the following will be described through specific embodiments.

[0053] Embodiment 1

[0054] Please refer to Figure 1 , which is a flowchart of a remote check-in method based on a large language model provided by the first embodiment of the present invention. The remote check-in method based on the large language model can be applied to any device or system. The remote check-in method based on the large language model includes the steps:

[0055] Step S10, obtain the user's check-in interaction information, and input the check-in interaction information into a pre-trained large prediction model for semantic analysis to obtain interaction semantics;

[0056] Among them, the remote check-in method based on the large language model is applied in a remote check-in terminal. The remote check-in terminal integrates three major interaction modules: voice, touch, and vision, providing users with a diversified interaction experience. The voice interaction module adopts an advanced speech recognition engine and is equipped with an efficient noise reduction algorithm to ensure that in an environment full of background noise, it can accurately capture and recognize the user's oral instructions and provide instant and clear voice feedback, enabling users to complete the check-in operation without hands. With a built-in translation engine, the terminal can intelligently switch languages to meet the needs of international users, making the terminal interaction more barrier-free and user-friendly.

[0057] The touch interaction module is based on a high-sensitivity capacitive screen and supports multi-touch and gesture operations. Users can easily browse flight information, select seats, etc. through an intuitive graphical user interface, greatly improving the convenience and intuitiveness of the operation. The vision interaction module realizes fast identity verification and personalized services through a high-definition camera and built-in face recognition technology.

[0058] The large language model built into the remote check-in terminal has powerful natural language processing capabilities and can accurately understand the complex instructions and intentions of users. It adopts a pre-trained language model and a domain-specific fine-tuning model, which can perform in-depth semantic parsing on the complex natural language input by users and accurately capture the intentions. In terms of dialogue management, the dialogue strategy is optimized through a reinforcement learning (RL) algorithm to ensure that the dialogue is coherent and logically consistent, supporting multi-round complex dialogue scenarios. The context memory function maintains the dialogue history through a long short-term memory (LSTM) network to improve the level of personalized services.

[0059] Optionally, in this embodiment, the large language model also supports multi-round conversations and continuous interactions, enabling more natural and fluent communication with users. Users can gradually clarify their needs during the conversation with the terminal and obtain a more personalized and considerate service experience. For example, when users are unsure how to fill in certain information, step-by-step guidance can be provided to ensure that each step proceeds smoothly. In addition, it can continuously optimize the conversation strategy based on the user's feedback to further improve the intelligent service level.

[0060] Optionally, before inputting the check-in interaction information into the pre-trained large prediction model for semantic analysis to obtain the interaction semantics, it further includes:

[0061] Obtain check-in interaction samples, and input the check-in interaction samples into the large prediction model for word embedding processing to obtain word embedding features; among them, by performing word embedding processing on the check-in interaction samples, the word embedding features of the check-in interaction samples can be effectively extracted;

[0062] Obtain the context features of the check-in interaction samples, and fuse the context features and the word embedding features to obtain fused features; among them, by fusing the context features and the word embedding features, the context information and the word embedding information can be effectively fused, improving the quality of feature extraction for the check-in interaction samples;

[0063] Perform semantic prediction on the fused features to obtain the predicted sample semantics, and determine the predicted sample reply according to the predicted sample semantics; among them, by performing semantic prediction on the fused features, the text semantics corresponding to the check-in interaction samples can be effectively predicted;

[0064] Determine the model loss according to the predicted sample reply, the predicted sample semantics, and the word embedding features, and update the large language model according to the model loss until the large language model converges to obtain the pre-trained large prediction model.

[0065] Furthermore, obtaining the context features of the check-in interaction samples includes:

[0066] Obtain the hidden state of the current text input in the check-in interaction sample and the hidden state of the text at the previous moment, and determine the forgetting coefficient according to the hidden state of the current text input and the hidden state of the text at the previous moment; among them, the hidden state contains the feature information of the corresponding text, and the forgetting coefficient is used to determine the text information that the corresponding text needs to forget;

[0067] Determine the forgotten information according to the forgetting coefficient and the memory state of the previous text, and perform a non-linear transformation on the hidden state of the current text input and the hidden state of the previous text to obtain an input coefficient; wherein, the input coefficient is used to determine the text information that needs to be added to the corresponding text.

[0068] Update the memory state of the current text input according to the input coefficient, and determine an output gate vector according to the hidden state of the current text input and the hidden state of the previous text.

[0069] Determine the context feature of the current text input according to the output gate vector and the updated memory state of the current text input.

[0070] Step S20, determine an interactive response according to the interactive semantics, and perform an interactive feedback to the user according to the interactive response.

[0071] Among them, calculate the semantic similarity between the interactive semantics and the preset semantics, determine the target semantics as the preset semantics corresponding to the maximum semantic similarity, obtain the response information corresponding to the target semantics in the interactive database, and obtain the interactive response. Different target semantics and corresponding response information are stored in the interactive database.

[0072] Step S30, if the interactive semantics contains a preset remote check-in semantics, obtain the travel destination and check-in requirements in the interactive semantics.

[0073] Among them, if the interactive semantics contains a preset remote check-in semantics, it is determined that the user has a remote check-in requirement. The preset remote check-in semantics can be set according to the requirements. By obtaining the travel destination and check-in requirements in the interactive semantics, it effectively facilitates the determination of subsequent check-in recommendation information.

[0074] Step S40, obtain the user's historical check-in information, and determine the check-in preference according to the historical check-in information.

[0075] Among them, the user's historical check-in information includes flight information of the user's historical flights, historical seats, historical waiting times, historical travel modes, accommodation and other information. Based on the historical check-in information, the user's flight preference, seat preference, travel preference and accommodation preference can be effectively determined, improving the accuracy of subsequent remote check-in recommendations for the user.

[0076] Optionally, determining the check-in preference according to the historical check-in information includes:

[0077] Obtain the historical flight and historical seat in the historical check-in information, and determine the flight preference according to the flight information of the historical flight; among them, obtain the flight model in the flight information, sort according to the number of each flight model to obtain the model sorting, and determine the flight preference based on the sorting order of the flight models in the model sorting;

[0078] Determine the seat preference according to the seat information of the historical seat. Among them, the check-in preference includes flight preference and seat preference. Generate a seat heat map based on the seat information, determine the seat area of interest based on the seat heat map, and obtain the seat preference.

[0079] Step S50, determine the check-in recommendation information according to the check-in preference, the travel destination and the check-in demand, and remotely recommend the check-in recommendation information to the user;

[0080] Among them, based on the knowledge graph in the aviation field, intelligent services such as flight information query, seat recommendation and luggage regulation explanation are provided. When the user asks "How many vacant seats are there on tomorrow's flight", it can not only query and display the vacant seat information, but also recommend the most suitable flight and seat options according to the user's check-in preference, and provide highly personalized suggestions and services for the user by combining historical data and real-time information.

[0081] Optionally, determining the check-in recommendation information according to the check-in preference, the travel destination and the check-in demand includes:

[0082] Obtain the departure date in the check-in demand, and determine the recommended flight number according to the departure date, the travel destination and the flight preference; among them, match the departure date, the travel destination and the flight preference with the flight data list to obtain the recommended flight number, and the flight data list stores the corresponding relationship between different departure dates, different travel destinations, different flight preferences and the corresponding recommended flight numbers;

[0083] Obtain the flight ticket sales information of the recommended flight number, and determine the recommended seat according to the flight ticket sales information and the seat preference; among them, match the remaining seats in the flight ticket sales information with the seats corresponding to the seat preference to obtain the recommended seat;

[0084] Generate the check-in recommendation information according to the flight information of the recommended flight number and the recommended seat.

[0085] Furthermore, after determining the recommended seat according to the flight ticket sales information and the seat preference, it further includes:

[0086] Obtain the historical travel mode and historical waiting time in the historical check-in information, and determine the travel preference according to the historical travel mode; wherein, obtain the means of transportation in the historical travel mode, determine the travel preference based on the number of times of the means of transportation, and the historical waiting time is the average value of the time difference between the boarding time and the time when the user arrives at the airport;

[0087] Obtain the current accommodation location of the user and the airport location corresponding to the recommended flight number, and determine the recommended departure time according to the travel preference, the current accommodation location, the airport location and the historical waiting time; wherein, calculate the travel time based on the travel preference, the current accommodation location and the airport location, and determine the recommended departure time based on the travel time, the historical waiting time and the latest boarding time corresponding to the recommended flight number;

[0088] Obtain the accommodation information in the historical check-in information, and determine the accommodation preference according to the accommodation information; wherein, the accommodation information includes information such as the type of accommodation hotel and the distance between the hotel and the airport;

[0089] Determine the accommodation recommendation according to the travel destination and the accommodation preference, and generate the check-in recommendation information according to the flight information of the recommended flight number, the recommended seat, the accommodation recommendation and the recommended waiting time.

[0090] Furthermore, after remotely recommending the check-in recommendation information to the user, it further includes:

[0091] If an artificial assistance instruction for the remote check-in recommendation is received, connect the user with the remote check-in customer service for audio and video connection, and obtain the remote check-in interaction information of the user within a preset duration; wherein, when the artificial assistance instruction is received, it is determined that the user needs artificial service, and by connecting the user with the remote check-in customer service for audio and video connection, the artificial consultation operation of the user is effectively facilitated;

[0092] Send the remote check-in interaction information and the check-in recommendation information to the remote check-in customer service, and obtain the artificial consultation questions of the user in real time; wherein, by sending the remote check-in interaction information and the check-in recommendation information to the remote check-in customer service, the remote check-in customer service's understanding of the user's needs is effectively facilitated;

[0093] Determine the question-and-answer reply based on the artificial consultation question and send the question-and-answer reply to the remote check-in customer service; in this embodiment, during the check-in process, the user only needs to input the identity verification information and establish an audio-video connection with the remote check-in customer service personnel through the one-key assistance module. Through the high-definition audio-video call function and screen sharing technology, the customer service personnel can guide the user to complete the check-in procedures in real time, including flight selection, seat confirmation, etc. During the entire call, the large model can analyze the user's questions in real time and extract relevant information from the knowledge base to provide instant answer suggestions for the customer service personnel. After the check-in is completed, a QR code voucher can be automatically generated and printing service is provided. The user only needs to carry this voucher and the identity document to board the plane easily.

[0094] Step S60, if the determination information for the remote check-in recommendation is received, then perform remote check-in registration for the user according to the remote check-in recommendation;

[0095] Among them, through the real-time docking with the airport flight interface, the remote check-in terminal provides users with detailed and up-to-date flight information query services. Users can easily view detailed information such as flight numbers, departure times, and arrival times on the terminal to better plan their trips.

[0096] In this embodiment, by inputting the check-in interaction information into the pre-trained large prediction model for semantic analysis, the interaction semantics of the user can be effectively obtained. Through the interaction semantics, the interaction reply can be effectively determined. Through the interaction reply, the user can be effectively interactively feedback. By obtaining the user's historical check-in information, the check-in preferences of the user can be effectively determined. Based on the check-in preferences, travel destinations, and check-in requirements, the check-in recommendation information can be automatically determined. Through the check-in recommendation information, the remote check-in recommendation for the user can be effectively based on the user's check-in habits, effectively facilitating the user's remote check-in operation and improving the user's experience.

[0097] Furthermore, through improvements and optimizations in multiple aspects such as constructing a perfect system architecture, upgrading function implementation, optimizing user experience, and enhancing security, the problems existing in the prior art in remote check-in have been successfully solved, not only improving the travel efficiency and user satisfaction, but also providing strong support for the intelligent transformation and sustainable development of the airport. Specifically as follows:

[0098] 1. By constructing a perfect remote check-in system, seamless docking with the airport information system, external service interfaces, and call center systems is achieved. This highly integrated system architecture not only ensures the real-time synchronization and update of flight information, but also greatly enriches the service content, providing users with a more convenient and comprehensive travel experience. Compared with the prior art, it no longer relies on a single information source or service channel, thus effectively avoiding the inconvenience to users caused by information lag or service shortage.

[0099] 2. Through a series of functions such as intelligent Q&A, one-key assistance calling, remote audio and video calls, flight inquiries, and remote check-in, users can complete the check-in procedures through intelligent terminals anytime and anywhere, without having to queue at the airport. This "online + offline" service model not only greatly improves travel efficiency but also significantly enhances user satisfaction and loyalty.

[0100] 3. In terms of user experience and security, advanced large language model technology and multi-modal interaction technology are introduced. The application of the large language model enables the terminal to more accurately understand the intentions and needs of users and provide intelligent and personalized services. At the same time, the multi-modal interaction technology enables the system to flexibly switch interaction methods and provide natural and smooth interaction experiences. In terms of security, through advanced encryption technology and security protection measures, combined with the intelligent analysis capabilities of the large model, potential security risks can be monitored and identified in real time to ensure the secure transmission and storage of user data.

[0101] Embodiment Two

[0102] Please refer to Figure 2 , which is a schematic structural diagram of a remote check-in system 100 based on a large language model provided by the second embodiment of the present invention, including:

[0103] A semantic analysis module 10, configured to obtain the user's check-in interaction information and input the check-in interaction information into a pre-trained large prediction model for semantic analysis to obtain interaction semantics.

[0104] Optionally, the semantic analysis module 10 is further configured to: obtain check-in interaction samples, and input the check-in interaction samples into the large prediction model for word embedding processing to obtain word embedding features;

[0105] obtain the context features of the check-in interaction samples, and fuse the context features and the word embedding features to obtain fused features;

[0106] perform semantic prediction on the fused features to obtain predicted sample semantics, and determine a predicted sample reply according to the predicted sample semantics;

[0107] determine a model loss according to the predicted sample reply, the predicted sample semantics, and the word embedding features, and update the large language model according to the model loss until the large language model converges to obtain the pre-trained large prediction model.

[0108] Furthermore, the semantic analysis module 10 is further configured to: obtain the hidden state of the current text input and the hidden state of the previous text in the check-in interaction samples, and determine a forgetting coefficient according to the hidden state of the current text input and the hidden state of the previous text.

[0109] Determine the forgotten information according to the forgetting coefficient and the memory state of the previous text, and perform a non-linear transformation on the hidden state of the current text input and the hidden state of the previous text to obtain the input coefficient;

[0110] Update the memory state of the current text input according to the input coefficient, and determine the output gate vector according to the hidden state of the current text input and the hidden state of the previous text;

[0111] Determine the context feature of the current text input according to the output gate vector and the updated memory state of the current text input.

[0112] The interaction feedback module 11 is used to determine an interaction reply according to the interaction semantics and perform interaction feedback on the user according to the interaction reply.

[0113] The requirement acquisition module 12 is used to obtain the travel destination and check-in requirements in the interaction semantics if the interaction semantics contains a preset remote check-in semantics.

[0114] The preference determination module 13 is used to obtain the user's historical check-in information and determine the check-in preference according to the historical check-in information.

[0115] Optionally, the preference determination module 13 is further used to: obtain the historical flight and historical seat in the historical check-in information, and determine the flight preference according to the flight information of the historical flight;

[0116] Determine the seat preference according to the seat information of the historical seat, where the check-in preference includes the flight preference and the seat preference.

[0117] The check-in recommendation module 14 is used to determine check-in recommendation information according to the check-in preference, the travel destination and the check-in requirements, and perform remote check-in recommendation on the user for the check-in recommendation information;

[0118] If a confirmation message for the remote check-in recommendation is received, perform remote check-in registration on the user according to the remote check-in recommendation.

[0119] Optionally, the check-in recommendation module 14 is further used to: obtain the departure date in the check-in requirements, and determine the recommended flight number according to the departure date, the travel destination and the flight preference;

[0120] Obtain the flight ticket sales information of the recommended flight number, and determine the recommended seat according to the flight ticket sales information and the seat preference;

[0121] Generate the check-in recommendation information based on the flight information of the recommended flight number and the recommended seat.

[0122] Further, the check-in recommendation module 14 is further configured to: obtain the historical travel mode and historical waiting time in the historical check-in information, and determine the travel preference according to the historical travel mode;

[0123] Obtain the current accommodation location of the user and the airport location corresponding to the recommended flight number, and determine the recommended departure time according to the travel preference, the current accommodation location, the airport location, and the historical waiting time;

[0124] Obtain the accommodation information in the historical check-in information, and determine the accommodation preference according to the accommodation information;

[0125] Determine the accommodation recommendation according to the travel destination and the accommodation preference, and generate the check-in recommendation information according to the flight information of the recommended flight number, the recommended seat, the accommodation recommendation, and the recommended waiting time.

[0126] Even further, the check-in recommendation module 14 is further configured to: if an artificial assistance instruction for the remote check-in recommendation is received, perform an audio-video connection between the user and the remote check-in customer service, and obtain the remote check-in interaction information of the user within a preset duration;

[0127] Send the remote check-in interaction information and the check-in recommendation information to the remote check-in customer service, and obtain the artificial consultation questions of the user in real time;

[0128] Determine the question and answer reply according to the artificial consultation question, and send the question and answer reply to the remote check-in customer service.

[0129] In this embodiment, by inputting the check-in interaction information into the pre-trained large prediction model for semantic analysis, the interaction semantics of the user can be effectively obtained. Through the interaction semantics, the interaction reply can be effectively determined. Through the interaction reply, the user can be effectively interactively feedback. By obtaining the historical check-in information of the user, the check-in preference of the user can be effectively determined. Based on the check-in preference, travel destination, and check-in requirements, the check-in recommendation information can be automatically determined. Through the check-in recommendation information, the remote check-in recommendation for the user can be effectively based on the user's check-in habits, effectively facilitating the user's remote check-in operation and improving the user's experience.

[0130] Embodiment III

[0131] Figure 3 It is a structural block diagram of a terminal device 2 provided in the third embodiment of the present application. As Figure 3As shown, the terminal device 2 of this embodiment includes: a processor 20, a memory 21, and a computer program 22 stored in the memory 21 and executable on the processor 20, such as a program for the remote check-in method based on a large language model. When the processor 20 executes the computer program 22, the steps in each of the above embodiments of the remote check-in method based on a large language model are implemented.

[0132] Exemplarily, the computer program 22 can be divided into one or more modules. The one or more modules are stored in the memory 21 and executed by the processor 20 to complete this application. The one or more modules can be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program 22 in the terminal device 2. The terminal device may include, but is not limited to, the processor 20 and the memory 21.

[0133] The so-called processor 20 may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0134] The memory 21 may be an internal storage unit of the terminal device 2, such as the hard disk or memory of the terminal device 2. The memory 21 may also be an external storage device of the terminal device 2, such as a plug-in hard disk equipped on the terminal device 2, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory 21 may also include both the internal storage unit and the external storage device of the terminal device 2. The memory 21 is used to store the computer program and other programs and data required by the terminal device. The memory 21 may also be used to temporarily store data that has been output or is to be output.

[0135] In addition, each functional module in various embodiments of the present application may be integrated into a processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of a software functional unit.

[0136] If the integrated module is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a computer-readable storage medium. Among them, the computer-readable storage medium may be non-volatile or volatile. Based on such an understanding, to implement all or part of the processes in the above-mentioned embodiment methods of the present application, it may also be completed by instructing relevant hardware through a computer program. The computer program may be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments may be implemented. Among them, the computer program includes computer program code, and the computer program code may be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable storage medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable storage medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0137] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A remote check-in method based on a large language model, characterized in that: The method comprises: Obtaining the user's check-in interaction information, and inputting the check-in interaction information into the pre-trained large prediction model for semantic analysis to obtain interaction semantics; Determining an interaction response according to the interaction semantics, and providing interaction feedback to the user according to the interaction response; If the interaction semantics includes preset remote check-in semantics, obtaining the travel destination and check-in requirements in the interaction semantics; Acquire historical check-in information of the user, and determine a check-in preference according to the historical check-in information; Determine check-in recommendation information according to the check-in preference, the travel destination and the check-in requirement, and make a remote check-in recommendation to the user based on the check-in recommendation information; If confirmation information for the remote check-in recommendation is received, remote check-in registration is performed for the user according to the remote check-in recommendation.

2. The remote check-in method based on a large language model as claimed in claim 1, characterized in that: Determining a check-in preference according to the historical check-in information includes: Acquire historical flights and historical seats in the historical check-in information, and determine flight preferences based on the flight information of the historical flights; A seat preference is determined according to the seat information of the historical seats, wherein the check-in preference includes the flight preference and the seat preference.

3. The remote check-in method based on a large language model as claimed in claim 2, characterized in that: Determining check-in recommendation information according to the check-in preference, the travel destination, and the check-in requirement, including: Obtaining the departure date in the check-in requirement, and determining a recommended flight number according to the departure date, the travel destination and the flight preference; Acquire flight ticket information of the recommended flight number, and determine a recommended seat according to the flight ticket information and the seat preference; The check-in recommendation information is generated according to the flight information of the recommended flight number and the recommended seat.

4. The remote check-in method based on a large language model as claimed in claim 2, characterized in that: After determining the recommended seat according to the flight ticket information and the seat preference, the method further includes: Acquire historical travel modes and historical waiting times in the historical check-in information, and determine travel preferences based on the historical travel modes; Obtaining the user's current accommodation location and the airport location corresponding to the recommended flight number, and determining a recommended departure time based on the travel preference, the current accommodation location, the airport location, and historical waiting time; Acquiring accommodation information from the historical check-in information, and determining accommodation preference according to the accommodation information; Accommodation recommendations are determined according to the travel destination and the accommodation preferences, and the check-in recommendation information is generated according to the flight information of the recommended flight number, the recommended seat, the accommodation recommendation, and the recommended waiting time.

5. The remote check-in method based on a large language model as claimed in claim 1, characterized in that: After recommending the check-in recommendation information to the user for remote check-in, the method further includes: If a manual assistance instruction recommended for the remote check-in is received, the user is connected to the remote check-in customer service via audio or video, and the remote check-in interaction information of the user within a preset time period is obtained; Sending the remote check-in interaction information and the check-in recommendation information to the remote check-in customer service, and obtaining the user's manual consultation questions in real time; Determine a question and answer response based on the manual consultation question, and send the question and answer response to the remote check-in customer service.

6. The remote check-in method based on a large language model as claimed in claim 1, characterized in that: The check-in interaction information is input into the pre-trained big prediction model for semantic analysis to obtain the interaction semantics, and the following is also included: Obtain a check-in interaction sample, and input the check-in interaction sample into the large prediction model for word embedding processing to obtain a word embedding feature; Acquire the context feature of the check-in interaction sample, and fuse the context feature with the word embedding feature to obtain a fused feature; Performing semantic prediction on the fused features to obtain predicted sample semantics, and determining predicted sample responses according to the predicted sample semantics; The model loss is determined according to the predicted sample response, the predicted sample semantics and the word embedding feature, and the large language model is updated according to the model loss until the large language model converges to obtain the pre-trained large prediction model.

7. The remote check-in method based on a large language model as claimed in claim 6, characterized in that: The context features of the check-in interaction sample are obtained, including: Obtaining the hidden state of the current text input and the hidden state of the text at the previous moment in the check-in interaction sample, and determining the forgetting coefficient according to the hidden state of the current text input and the hidden state of the text at the previous moment; Determine the forgetting information according to the forgetting coefficient and the memory state of the text at the previous moment, and perform nonlinear transformation on the hidden state of the current text input and the hidden state of the text at the previous moment to obtain the input coefficient; The memory state of the current text input is updated according to the input coefficient, and the output gate vector is determined according to the hidden state of the current text input and the hidden state of the text at the previous moment; The context feature of the current text input is determined according to the output gate vector and the updated memory state of the current text input.

8. A remote check-in system based on a large language model, characterized in that: The system comprises: A semantic analysis module is used to obtain the user's check-in interaction information, and input the check-in interaction information into the pre-trained large prediction model for semantic analysis to obtain interaction semantics; An interaction feedback module, used to determine an interaction reply according to the interaction semantics, and provide interaction feedback to the user according to the interaction reply; A demand acquisition module, configured to acquire the travel destination and check-in demand in the interaction semantics if the interaction semantics includes preset remote check-in semantics; a preference determination module, configured to obtain the historical check-in information of the user and determine the check-in preference according to the historical check-in information; A check-in recommendation module, configured to determine check-in recommendation information according to the check-in preference, the travel destination and the check-in requirement, and perform remote check-in recommendation to the user based on the check-in recommendation information; If confirmation information for the remote check-in recommendation is received, remote check-in registration is performed for the user according to the remote check-in recommendation.

9. The remote check-in system based on a large language model as claimed in claim 8, characterized in that: The preference determination module is also used to: Acquire historical flights and historical seats in the historical check-in information, and determine flight preferences based on the flight information of the historical flights; A seat preference is determined according to the seat information of the historical seats, wherein the check-in preference includes the flight preference and the seat preference.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images